diff --git a/pyproject.toml b/pyproject.toml
index 896465bec..c4c594a29 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -12,10 +12,30 @@ packages = [{ include = "ragstack" }]
 [tool.poetry.dependencies]
 python = ">=3.9,<3.12"
 astrapy = "~0.7.0"
-cassio = "~0.1.3"
-unstructured = { version = "0.12.5" }
-llama-index = { version = "0.9.48", extras = ["langchain"] }
-llama-parse = { version = "0.1.4" }
+cassio = "~0.1.4"
+unstructured = "0.12.5"
+
+# llama-index
+llama-index = "0.10.19"
+llama-index-vector-stores-astra-db = "0.1.4"
+llama-index-vector-stores-cassandra = "0.1.3"
+llama-index-embeddings-langchain = "0.1.2"
+llama-parse = "0.3.8"
+# optional integrations
+## azure
+llama-index-llms-azure-openai = { version = "0.1.5", optional = true }
+llama-index-embeddings-azure-openai = { version = "0.1.6", optional = true }
+## bedrock
+llama-index-llms-bedrock = { version = "0.1.3", optional = true }
+llama-index-embeddings-bedrock = { version = "0.1.4", optional = true }
+## google
+llama-index-llms-gemini = { version = "0.1.5", optional = true }
+llama-index-multi-modal-llms-gemini = { version = "0.1.3", optional = true }
+llama-index-llms-vertex = { version = "0.1.4", optional = true }
+llama-index-embeddings-gemini = { version = "0.1.4", optional = true }
+
+
+# langchain
 langchain = "0.1.12"
 langchain-core = "0.1.31"
 langchain-community = "0.0.28"
@@ -28,6 +48,11 @@ langchain-nvidia-ai-endpoints = { version = "0.0.3", optional = true }
 langchain-google = ["langchain-google-genai", "langchain-google-vertexai"]
 langchain-nvidia = ["langchain-nvidia-ai-endpoints"]
 
+llama-index-google = ["llama-index-llms-gemini", "llama-index-multi-modal-llms-gemini", "llama-index-llms-vertex", "llama-index-embeddings-gemini"]
+llama-index-azure = ["llama-index-llms-azure-openai", "llama-index-embeddings-azure-openai"]
+llama-index-bedrock = ["llama-index-llms-bedrock", "llama-index-embeddings-bedrock"]
+
+
 [tool.poetry.group.test.dependencies]
 pytest = "*"
 nbmake = "*"
diff --git a/ragstack-e2e-tests/pyproject.langchain.toml b/ragstack-e2e-tests/pyproject.langchain.toml
index e3c0cdfe1..7685d1ef6 100644
--- a/ragstack-e2e-tests/pyproject.langchain.toml
+++ b/ragstack-e2e-tests/pyproject.langchain.toml
@@ -13,7 +13,6 @@ pytest = "*"
 black = "*"
 ruff = "*"
 google-cloud-aiplatform = "^1.36.4"
-boto3 = "^1.29.6"
 huggingface-hub = "^0.20.3"
 azure-storage-blob = "^12.19.0"
 pillow = "^10.2.0"
@@ -33,12 +32,25 @@ langchain-openai = { git = "https://github.com/langchain-ai/langchain.git", bran
 langchain-google-genai = { git = "https://github.com/langchain-ai/langchain-google.git", branch = "main", subdirectory = "libs/genai" }
 langchain-google-vertexai = { git = "https://github.com/langchain-ai/langchain-google.git", branch = "main", subdirectory = "libs/vertexai" }
 langchain-nvidia-ai-endpoints = { git = "https://github.com/langchain-ai/langchain-nvidia.git", branch = "main", subdirectory = "libs/ai-endpoints" }
-llama-index = { version = "0.9.48", extras = ["langchain"] }
-llama-parse = { version = "0.1.4" }
+
+llama-index = "0.10.18"
+llama-index-vector-stores-astra-db = "0.1.4"
+llama-index-vector-stores-cassandra = "0.1.3"
+llama-index-llms-azure-openai = "0.1.5"
+llama-index-embeddings-azure-openai = "0.1.6"
+llama-index-llms-bedrock = "0.1.3"
+llama-index-embeddings-bedrock = "0.1.4"
+llama-index-llms-gemini = "0.1.5"
+llama-index-multi-modal-llms-gemini = "0.1.3"
+llama-index-llms-vertex = "0.1.4"
+llama-index-embeddings-gemini = "0.1.4"
+llama-index-embeddings-huggingface = "^0.1.0"
+llama-index-llms-huggingface = "^0.1.0"
+llama-parse = "0.3.8"
+
 astrapy = "~0.7.0"
-# we need this specific feature from cassio: https://github.com/CassioML/cassio/pull/128
 cassio = "~0.1.4"
-unstructured = { git = "https://github.com/Unstructured-IO/unstructured.git", branch = "main" }
+unstructured = "0.12.5"
 
 [build-system]
 requires = ["poetry-core"]
diff --git a/ragstack-e2e-tests/pyproject.llamaindex.toml b/ragstack-e2e-tests/pyproject.llamaindex.toml
index 4122bf621..52aafe831 100644
--- a/ragstack-e2e-tests/pyproject.llamaindex.toml
+++ b/ragstack-e2e-tests/pyproject.llamaindex.toml
@@ -12,8 +12,6 @@ python = ">=3.9,<3.12,!=3.9.7"
 pytest = "*"
 black = "*"
 ruff = "*"
-google-cloud-aiplatform = "^1.36.4"
-boto3 = "^1.29.6"
 huggingface-hub = "^0.20.3"
 azure-storage-blob = "^12.19.0"
 pillow = "^10.2.0"
diff --git a/ragstack-e2e-tests/pyproject.toml b/ragstack-e2e-tests/pyproject.toml
index be6bf4f1a..b058b354e 100644
--- a/ragstack-e2e-tests/pyproject.toml
+++ b/ragstack-e2e-tests/pyproject.toml
@@ -13,8 +13,8 @@ python = ">=3.9,<3.12,!=3.9.7"
 pytest = "*"
 black = "*"
 ruff = "*"
-boto3 = "^1.34.21"
-huggingface-hub = "^0.20.3"
+llama-index-embeddings-huggingface = "^0.1.0"
+llama-index-llms-huggingface = "^0.1.0"
 azure-storage-blob = "^12.19.0"
 pillow = "^10.2.0"
 testcontainers = "^3.7.1"
@@ -24,9 +24,7 @@ trulens-eval = "^0.21.0"
 # From LangChain optional deps, needed by WebBaseLoader
 beautifulsoup4 = "^4"
 
-ragstack-ai = { path = "../", develop = false, extras = ["langchain-google", "langchain-nvidia"]}
-# we need this specific feature from cassio: https://github.com/CassioML/cassio/pull/128
-cassio = "~0.1.4"
+ragstack-ai = { path = "../", develop = false, extras = ["langchain-google", "langchain-nvidia", "llama-index-google", "llama-index-azure", "llama-index-bedrock"]}
 
 # benchmarks
 pyperf = "^2.6.2"
diff --git a/tests/unit-tests/test_ragstack.py b/tests/unit-tests/test_ragstack.py
index b10b4dd1a..9238ebe4b 100644
--- a/tests/unit-tests/test_ragstack.py
+++ b/tests/unit-tests/test_ragstack.py
@@ -1,6 +1,6 @@
 def test_import():
-    from llama_index.vector_stores import AstraDBVectorStore  # noqa
-    from llama_index.vector_stores import CassandraVectorStore  # noqa
+    from llama_index.vector_stores.astra_db import AstraDBVectorStore  # noqa
+    from llama_index.vector_stores.cassandra import CassandraVectorStore  # noqa
     from langchain.vectorstores import AstraDB  # noqa
     import langchain_core  # noqa
     import langsmith  # noqa